Being able to correctly characterise an observed time series into its separate difference stationary and trend stationary regimes, should they exist, has important implications for effective model building and forecasting in economics and finance. Existing ratio-based statistics test the null hypothesis that a time series displays constant trend stationarity, I(0), against the alternative of a change in persistence from trend stationarity to difference stationarity, I(1), or vice versa. Here, however, we demonstrate that these tests are unable to adequately discern between a true change in persistence and a constant I(1) process. We propose modified tests which, by design, have the same critical values regardless of whether the process is I(0) or I(1) throughout. Hence, our null hypothesis is that of constant persistence (either constant I(0) or constant I(1)). Tests directed against both I(1) to I(0) and I(0) to I(1) persistence changes are provided, together with tests where the direction of change under the alternative is unspecified. Our tests retain the same rates of consistency against persistence change processes as their unmodified counterparts. Simulation evidence suggests that our new procedures work extremely well in practice, with the modified tests correctly being sized in both constant I(0) and constant I(1) environments, and displaying only very modest losses in power, relative to unmodified tests, against persistence change process
